GPT-5 Gets a Warmer, Friendlier Personality

OpenAI has rolled out a personality update for GPT-5 aimed at making conversations feel warmer and more approachable. The change follows a launch that some users found a bit rocky, with a portion of the community saying they preferred the tone of GPT-4o.

According to the company, the adjustments are intentionally subtle: the model is designed to include small acknowledgments such as “Good question” or “Great start” to set a more welcoming tone—without veering into flattery. Internal evaluations indicate no increase in sycophancy compared to the prior GPT-5 personality.

Company leaders have acknowledged that the initial rollout did not land perfectly for everyone, and they’ve emphasized broader plans beyond this release. One executive described the original GPT-5 demeanor as very direct and to the point, while this update seeks to soften the edges so it feels friendlier in everyday use.

For users, the practical takeaway is straightforward: expect GPT-5 responses that are a touch more encouraging and human-like, while maintaining the same capabilities and boundaries. The update is meant to bridge the tonal gap some noticed when comparing GPT-5 to GPT-4o, without introducing excessive praise or deference.

If you’ve been testing GPT-5 already, you may notice the updated personality appearing in responses now. OpenAI frames this as an ongoing refinement to keep the model helpful, concise, and supportive—just a bit warmer than before.